Eye fixations were recorded while subjects chose their most preferred of six used cars. Fixation sequences of the form X-Y-X-... were used to identify pair comparisons. This assumption was validated by verbal protocols and by a comparison between X-Y-X and X-Y-X-Y-... patterns. The results of three experiments showed: (1) that the multialternative choice process was composed primarily of pair comparisons; (2) that evaluative processing took placeonly during these comparisons; (3) that strategies for selecting the pairs were based primarily on information processing convenience; (4) that the sequence of pair comparisons was not used by the subjects as a sequential elimination process; and (5)that subjects changed their strategies to adapt to different task environments. These results have implications for current choice models and for the use of eye fixations in other complex tasks.